The project builds on a prior app that was designed for Green Light Optimized Speed Advisory (GLOSA). This is more colloquially known as keeping a vehicle in the green wave: you are at a location and moving at a speed that will allow you to (theoretically) have a green light at each intersection you encounter along a corridor. Our long-term goal is to extend the FastTrack app described in the Background section to include actuated signals along a corridor. This project takes a first step by evaluating the effectiveness of machine-learning algorithms to predict the next phase of an actuated signal on a busy bike corridor, given information about the past K phases. In essence, this is what is called a time-series forecasting problem. If we find forecasting success here, then we can begin to incorporate these algorithms into a more comprehensive GLOSA app (post-grant).
The project used data captured during the prior V2X project captured from the 18th and Alder intersection in Eugene OR during 12 days in June 2018. Loop detectors and advanced loop detectors currently exist in both directions on Alder to recognize the presence of bicycles and vehicles. The dataset consists of phase-change data for a complicated intersection that plays a key role in a bike corridor. The intersection has eight separate phases, all callable, that serve vehicles, bicyclists, pedestrians, and buses, all in various combinations. The data was taken from the month of June 2018. June is typically a heavy biking month near the University of Oregon campus.
This work is marked with CC0 1.0 Universal
Fickas, Stephen. Data File: Green Waves, Machine Learning, and Predictive Analytics: Making Streets Better for People on Bikes. NITC-1299. Portland, OR: Transportation Research and Education Center (TREC), 2021. https://doi.org/10.15760/TREC_datasets.15